/sci/ - Science & Math

>>14876000
The algorithm could train itself to figure out the smallest maid and draw it for you if you trained it for a while with bigger ones I'm assuming. You'd just say "yeah that thing there? do it to these dimensions instead" or something.

For the record, I have no idea what I'm talking about. I thought the word "automaida" was funny, and so I'm responding.

>>14876009
I don't know if I can make a cellular automata that draws maids or not. The Wolfy book about how these apply to engineering won't be here for like a week. All I know is they draw stuff and Java is very good at drawing stuff.

I know for sure that I don't know how to make an AI capable of advanced image processing though.

>For the record, I have no idea what I'm talking about. I thought the word "automaida" was funny, and so I'm responding.

This is the power of putting anime maids in the sciences.

Here is one of Wolfy's talks about the things she made her computer draw.

>draw stuff with computer
https://youtu.be/_eC14GonZnU

The algorithms are not hard to make but they can draw bizarre things when repeated.

>At 3x3 I can't draw a maid
Your assumption is incorrect. You can draw a maid or whatever even on a 1x1 "grid", it just will not be recognizable. There is no exact resolution at which a maid becomes a maid, there are only degrees of maidness.

>>14876019
I guess at those sizes my problem is that the representation of the maid is too abstract. Most people when shown a single black square or a pattern in a 3x3 grid won't assume they are looking at an anime maid.

I wonder at what degree of maidness would most people agree that they are looking at an anime maid?

>>14876024
the recognizability of a picture depends on the autism of the viewer, take the 3x3 picrel for example...

>>14876027
forgot pic

>>14876028
How is this a maid? All I can see is 9 boxes.

>>14876024
Now the problem becomes "somehow measuring degrees of something being something". That's not an easy one. I guess you could arrive to an optimal resolution by making various drawings of maids in different resolutions and asking people to rate how maid-like a given picture is.

>>14876000
Current state of /sci/
for god's sake kill yourself

>>14876051
He talks about science with an anime twist on an anime website's science board.

>>14876051
at least it's not a trap...

>>14876031
it's not a maid and you should be thankful that you can't recognize what is on that 3x3 grid

>>14876000
>Somewhere between 100x100 and 3x3 must be the smallest possible maid.
That's not how human perception works. Look at picrel. Do you recognize what it's supposed to depict? Now show it to some rural 90 years old villager. Think they'll figure it out? This stuff is 90% convention. Come up with some convention for your maids that will be memorable to the viewer and you're done.

>>14876036
This is a pretty cool problem. I was also considering generating every possible grid pattern at various sizes and then looking through the output images to see which configurations represent maids.

If I generate every possible 100x100 grid then a subset of that data will be every possible 100x100 anime maid. They all live in that computation because every possible black and white arrangement of pixels lives in that computation.

To make all the images I would have to tell my computer to draw 2^10000 images. That's probably a big number. The science computer only has 1TB hard drive and that might be more than 1TB of images.

>>14876086
This makes a lot of sense. Thank you for the advice, fren.

>>14876000
try a n-dimensional CA - and then think about how you can quantize how much you care about molecular fidelity and how quickly you could simulate proteins then, if you only care about good-path outcomes

>>14876133
>2^10000 images
That's the number of images with black and white pixels only, if you're planning to generate every 100x100 grid with 256*256*256=16777216 RGB colors then that would take 16777216^10000 which is approximately 10^72247 this is many orders of magnitude more than the number of particles in the observable universe

>>14876000
PLEASE GO BACK TO /g/ AND TAKE YOUR HENTAI THREADS WITH YOU. THIS IS A SCIENCE BOARD WE DON'T JACK OFF HERE. AAAAAAAAA

>>14876150
Woah. Computing stuff is cool. I will try make an algorithm that filters some useless computations. For example on a 100x100 grid, if you only drew one black pixel, that's not a maid so there is no reason to search those images for maids.

If you drew 9,999 black pixels and only one white one there is no reason to search there either.

So somewhere there is both a minimum and maximum number of pixels which can be in a maid.

We can sharpen our maidseach algorithm by determining which sets of data can be ignored instead of computed. It may not be possible to get exact rules and I would rather err on the side of computing too much stuff than risk missing any maids. We can definitely eliminate some computations though.

>>14876170
Also it seems like it would be easy to parallelize this. Like one computer can generate images in a given range and another computer can generate images in a different range. This problem is vulnerable to multi-threading too because I don't need any other maid to try computing a new one.

I have two computers. The science computer and the DevTerm. The DevTerm is not operating right now because I can't figure out how to get it's OS to talk to Ubuntu repositories so I can install stuff on it.

The science computer I think maybe has a GPU in it? I have never tried to use the GPU if it does. It is a GPD Win 3 (all my code and acreenshots come from that machine).

I can probably search a lot of maids with two computers.

>>14876000
the old pokemon gameboy games had an anime maid sprite i think. something like 8x8 or 10x10

generating every possible 100x100 maid is not going to be a computationally achievable problem. you could start on the small end and go from there, looking at every possible binary 3x3 image is only 512 images for instance. youd start to see recognizable maids well before 100x100

>>14876195
This is a good idea. I didn't find the Gameboy maid specifically, but when I searched maid sprites there were definitely recognizable maids on much smaller grids than 100x100.

Every possible anime maid is going to have a subset which is every possible Tohru (see attachment). I wonder if I can get in copyright trouble for algorithmically generating a maid someone else already imagined and drew by hand?

>>14876141
I don't know what most of this means, but I like your energy. I probably won't go past 3 dimensions because I don't care about computing hyperdimensional maids which I can't visualize. I have to start at 2 dimensions though. There are no 1D maids.

>>14876159
I am also posting on /g/ in /dpt/. Dra/g/ons are talking about Forth in there and I like Forth.

If you want to participate in talking to dra/g/ons about Forth, please be nice on their board. I have attached a small etiquette guide for using the Dra/g/on Maid board.

>>>/g/88818660

JANNNNIEEEEESSSS

please go the fuck back. the jannies have thankfully cleaned up every thread you make here. i hope they continue to remove your retarded filth from this board

>>14876362
If you don't like anime and you don't like science, then why are you on the science board of an anime site? At any given time 50% of this board is /pol/-lite which is usually neither related to science nor anime. Why don't you complain about those threads instead? One guy trying to compute all the possible maids on a grid is higher quality and more on topic than any of the /pol/ threads.

>>14876383
im good thanks. i gain an immense amount of pleasure from doing this. just gonna keep bumping and filling up this and every other retarded anime maid thread i see on /sci/ until the jannies purge it.

oh, and if you want to keep defending someone who has themselves said on multiple occasions that they don't care to learn about math or science, you should go back to /g/ yourself and post your retarded anime maids there or anywhere but /sci/ for all i care.

>>14876000
Okay I think I made up an algorithm to compute every maid in a grid. All you have to do is count in binary. So consider a 4 bit number.

0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
1101
1110
1111

So I can count to 16 doing that. Just put a line break half way through the numbers to get the grids. This means I can compute the grids at the speed of counting:

00
00

00
01

00
10

00
11

01
00

01
01

01
10

01
11

10
00

10
01

10
10

10
11

11
00

11
01

11
10

11
11

Also we can show that 2x2 grids contain no maids, by exhaustion.

>>14876698
>>>/x/

>>14876736
Counting is paranormal?

>>14876000
XY = Man
OP = XY
OP = Man
That drawing = isn´t OP

Please Stop Posting Sexy OP Images, I Already Jerked Off Today And I'll Break My Dick If I Do It Again

>>14876858
Thank you fren, this is maximumly based and I will use the information to make the integrator better.

Can we agree to ban maidanon unless “he” actually shares some code? I suspect this is an elaborate larp.

>>14876000
> maid named Wolfy who likes cellular automata
infinite keks at wolfram
God damn it I will never stop laughing at this
This is the best thread on /sci/ in idk, 10 fucking years

>>14876056
lmao sharp radar you have there

Best thread on /sci/

>>14876000
This paper may be what you’re looking for:

https://distill.pub/2020/growing-ca/

>>14878573
This is cool. I like the self-regenerating indestructible lizard. Thank you for showing it to me.